Why Do Women Enter and Exit From Material Hardship?

Abstract

Current critiques of the official federal poverty measure have led to a growing interest in alternative measures of well-being, such as individual reports of the inability to meet basic needs. This article considers a wide set of risk factors for four different forms of material hardship (food insufficiency, utility disconnection, unmet medical needs, and housing problems) using data from a panel study of single mothers. Specifically, the authors analyze the role of maternal health, household composition, and income on entrances into and exits from material hardship. The results show that there is a great deal of heterogeneity across forms of material hardship but that in general, the predictions of the authors’ conceptual model are confirmed. Determinants of entrances into material hardship differ from those that predict exits, suggesting that interventions to help families exit from material hardship may need to address different issues than those that triggered the entrance into material hardship in the first place.

Keywords

poverty material hardship well-being food insufficiency welfare

Material hardship, reports of an inability to meet basic needs in areas such as food, housing, and medical care, is an area of growing research and policy interest (Beverly, 2000; Blank, 2007; Boushey & Gundersen, 2001; Ouellette, Burstein, Long, & Beecroft, 2004). While prevalence rates of different forms of material hardship and their correlates are well known for both the general population as well as women on welfare, much less is known about what triggers a change in material hardship status.¹ Factors associated with entering hardship may well differ dramatically from factors that are associated with exiting hardship and a return to equilibrium. Given the mounting evidence of the existence of negative consequences associated with experiencing material hardship for both adult health (Siefert, Bowman, Heflin, Danziger, & Williams, 2000) as well as parental stress and child development (Alaimo, Briefel, Frongillo, & Olson, 1998; Gershoff, Aber, Raver, & Lennon, 2007; Gundersen, Lohman, Garasky, Stewart, & Eisenmann, 2008), an understanding of the process through which households enter and exit material hardship is critical.

Furthermore, recent research indicates that different forms of material hardship, such as food hardship, housing problems, medical hardships, and bill-paying hardships, represent distinct latent constructs and cannot be simplified into a unidimensional construction of hardship without a serious loss of information (Heflin, Sandberg, & Rafail, 2009). An important implication is that determinants of change in material hardship status may differ across domains of material hardship. Additional support for the need to examine different domains of material hardship separately comes from qualitative research demonstrating that the work women do to avoid or remedy hardship varies by hardship type (Heflin, London, & Scott, 2011).

This article examines transitions into and exits from four different forms of material hardship (food insufficiency, utility disconnection, unmet medical needs, and housing problems) using a unique longitudinal data set, the Women’s Employment Survey. The aim of the project is twofold: First, we seek to identify if the risk factors for transitions in material hardship vary across hardship type. As a result of the richness of the data set, we are able to include measures of individual health not usually observed, such as depression, drug dependence, alcohol dependence, post-traumatic stress disorder, and domestic violence. Second, we separate transitions into hardship from transitions out of hardship to test our conceptual model that some factors are more closely associated with entry into material hardship than exits out.

This research has important implications for public policy. First, while changes in employment status and poverty status are well-researched topics, little is known about the predictors of change in material hardship status. This is surprising given the mounting critiques of the official federal poverty measure (for a recent review, see Blank, 2007) and evidence indicating that poverty and material hardship are related but distinct measures of economic well-being (Heflin et al., 2009). Therefore, understanding the determinants of change in material hardship, and how determinants differ across hardship measures, is of critical importance to the research community. Given the rising criticism of the income poverty measure, research documenting the social processes underlying experiences of material hardship is of practical policy importance. In particular, a better understanding of individual risk factors for transitions into material hardship as well as protective factors that are associated with transitions out of material hardship could help us better target limited public funds toward those at highest risk of different forms of material hardship as well as how to remedy hardship when it does occur. Screening tools to identify households at risk of material hardship are needed as well as a better understanding of the barriers households face in returning to equilibrium.

Below we review the literature on determinants of material hardship, most of which uses cross-sectional data. Then, we provide a conceptual framework for the proposed relationship between maternal health, household composition, and income and our four measures of material hardship. After describing the data and measures used, we explain our empirical strategy. Our results section details our findings for determinants of entry into four forms of material hardship and exit. We conclude by discussing patterns of results across hardship types and implications for both public policy and research.

Background Literature

Previous Estimates of Material Hardship

Research on material hardship has been motivated by concerns that the official federal poverty line is “broken.” Common critiques tend to focus on measurement issues in the federal poverty threshold (Citro & Michael, 1995; Rector, 1999; Sen, 1999). One set of suggestions is to change the calculation of family income to include the value of public transfers, such as food stamps, Medicaid, and housing subsidies, and to exclude some types of expenses, such as medical, child care, and work-related expenses, as well as taxes (Citro & Michael, 1995). Other suggestions focus on the need to revise the federal threshold itself and to possibly tie it to “averaged” family expenditures (Citro & Michael, 1995, p. 45). As a result of the criticism of the income poverty measure, there is a growing interest in using measures of material hardship to identify individuals who do not consume minimal levels of basic goods and services, such as food, housing, clothing, and medical care (Beverly, 2001; Boushey & Gundersen, 2001; Ouellette et al., 2004).

Given the growing interest in alternative measures of well-being, surprisingly little is known about the determinants of changes in material hardship status over time—almost all prior research has been conducted with cross-sectional data.² Cross-sectional estimates of the prevalence of different domains of material hardship vary depending on the data source and population studied.³ Heflin (2006) using the same data as that used in this study, the Women’s Employment Study (WES), finds substantial movement into and out of material hardship categories. Close to half the sample report an unmet medical need, food insufficiency, or poor housing at least once over the observation period; approximately 30% report a utility disconnection. Most of the population reporting a given hardship does so once or twice over the observation period. Therefore, given the large fraction of the population that changes status across domains of material hardship, it is possible to examine determinants of change in material hardship in a longitudinal framework.

Relatively few studies have examined determinants of material hardship using longitudinal data. Gundersen and Gruber (2001) and Heflin, Corcoran, and Siefert (2007) estimate determinants of food insufficiency with panel data. Results are generally quite mixed across the studies with regard to the importance of employment, income, and education as determinants of food insufficiency. In the study closest to this one, Heflin et al. use fixed-effect models to show that changes in food insufficiency status are related to changes in mental health status and level of income. Similar results are found in a longitudinal study of food insecurity among American children (Bhargava, Jolliffe, & Howard, 2008). An important exception to the focus on food insufficiency is found in the work of Kalil, Seefeldt, and Wang (2002) and Reichman, Teitler, and Curtis (2005), who both examine the effect of sanctions (for noncompliance with work and reporting requirements) on levels of material hardship among samples of welfare mothers. Kalil et al. (2002) find that being sanctioned, having a mental health problem, or a learning disability is associated with utility disconnections.⁴ Similarly, Reichman et al. (2005) report that being sanctioned increases the risk of hunger, utility shutoffs, and having an unmet medical need.⁵

Thus, while there is some evidence from panel studies that mental health and income are associated with food insufficiency, less is known about determinants of change (beyond being sanctioned) in other forms of material hardship status. While household food insufficiency remains an especially important form of material hardship because of the relatively high levels reported and possible consequences, other forms of material hardship could have important implications. For example, having unmet medical needs or living in hazardous housing conditions could also have negative health implications. Furthermore, it would be helpful to understand if the characteristics associated with entry into food insufficiency differ from those associated with other forms of material hardship as well as exits from hardship.

Finally, by separately examining transitions into and out of material hardship, a greater understanding of the processes at work may be gleaned. One limitation of the fixed-effect model is that the magnitude and signs for determinants of transitions into material hardship (coded as 0 for no and 1 for yes) are averaged across transitions into material hardship (0/1) and transitions out of material hardship (1/0). In reality, there are two very different processes at work and therefore two different models that should be considered. The first model examines all those who experience a transition into material hardship, conditional on not experiencing material hardship in the prior time period. The second model examines all those who are observed to leave material hardship, conditional on experiencing material hardship in the prior time period. By separating determinants of entry from those of exit, we are able to present a deeper description of the family processes at work as low-income families strive to meet their basic needs.

Conceptual Model

This project will examine changes in four forms of material well-being: food insufficiency status, utility disconnection, unmet medical needs, and housing problems. The four domains include items that can be categorized as both “critical” and “serious” hardships (Beverly, 2001; Boushey & Gundersen, 2001). Boushey and Gundersen (2001) contrast the two as follows:

Critical hardships explain the extent to which families fail to meet their basic needs for survival. In comparison, serious hardships explain the extent to which families lack the goods, services, and financial ability to maintain employment and a stable, healthy home environment. (p. 20)

In particular, food insufficiency, unmet medical needs, and utility disconnections can be classified as critical hardships. The remaining item, housing problems, can be considered a serious hardship. Heflin et al. (2009) have shown using national representative data that each of these four forms of material hardship represent distinct latent constructs and, although correlated, cannot be reduced. In this section, we discuss our conceptual framework used to construct separate models of transitions into and out of four forms of material hardship.

Income

All the four forms of hardship explored here indicate that financial resources are at least in some way insufficient to cover basic needs. However, the empirical literature suggests that income poverty and several key measures of material hardship are only moderately associated in the United States (Beverly, 2000; Boushey and Gundersen, 2001; Iceland & Bauman, 2004; Mayer, 1997; Mayer & Jencks, 1989, 1993; Rector, 1999). On the one hand, poor people are more likely than nonpoor people to report a variety of material hardships (Boushey et al., 2001). On the other hand, many poor people do not report some types of material hardship, and some people who are not poor do. One of the best-developed measures of material hardship, the Food Security Scale, correlates with income and poverty at approximately 0.33 (Hamilton et al., 1997), and a composite measure of material hardship based on the original index developed by Mayer and Jencks (1989) correlates with the official federal income poverty threshold at 0.18 (Short, 2005).⁶ Thus, empirical evidence would suggest that income alone is not fully determinative of material hardship.

Our expectation is that drops in income will be related to entrances into food insufficiency, utility hardships, and medical hardships but not housing hardships, which are less likely to be tied directly to short-term reductions in the income level of residents. Exits in hardship are expected to be associated with increases in income for all four types of hardship. In the case of housing hardships, an increase in income would be necessary to both remedy the hardship or to move. However, the threshold level of income necessary to enter or exit hardship could well differ across domains. Food insufficiency requires relatively small changes in income to remedy while the costs of turning on disconnected utilities or obtaining care from a doctor or dentist are likely to be much larger.⁷

Health

There are several mechanisms that render health status an important determinant of entrances and exits from material hardship.⁸ First, individual circumstances, such as mental illness, introduce variation in coping mechanisms for dealing with scarce resources that may impede or distort budgeting and planning behavior.⁹ Second, health problems may create competing demands that households may choose to maximize at the expense of some forms of material hardship. Health care expenses are an excellent example of a good that households might place above necessities, including food. Although the prior literature focuses on food insufficiency alone (Heflin et al., 2007; Nelson, Brown, & Lurie, 1998), our expectation is that an increase in the presence of physical and mental health problems will also be associated with entrances into medical hardships and utility disconnections and, given the possibility that health problems may lead to changes in residence, housing hardships. It is less clear if the absence of health problems among those who are already experiencing material hardship will increase exits from material hardship, holding income constant.

Household Composition

There are several reasons to believe that changes in household composition may influence reports of material hardship. To the extent that additional adults are present in the household, if they are employed and share their income with the household, this may lessen reports of material hardship. However, if they do not work and/or if they have health conditions that require additional expenditures, additional adults may increase levels of material hardship, all else equal. Similarly, the number of children is likely to increase total demands that could lead to an increase in material hardship. However, it is unclear if households would maximize some forms of material well-being over others in order to protect their children from some forms of hardship (Polit, London, & Martinez, 2000), suggesting that the effect may not be uniform across hardship type.

In summary, based on prior research, we employ a conceptual model that changes in income, health, and household composition are correlated with changes in material hardship in a manner that is not likely to be equivalent across the different forms of material hardship or even for entrances to or exits from the same form of hardship in all cases. In the next section, we explain how we operationalize this conceptual model.

Data and Method

Data

In this study, we will analyze data from the five waves of the WES, a panel survey of barriers to employment among 753 mothers who were receiving cash assistance in an urban Michigan county in February 1997. Trained staff of the Survey Research Center of the Institute for Social Research of the University of Michigan conducted face-to-face, in-home, structured interviews between August and December of 1997, August and December 1998, November 1999 and March 2000, September and December 2001, and November 2003 and March 2004.¹⁰ The first two interviews lasted approximately 1 hour; the third, about 90 minutes; and the fourth and fifth, about 85 minutes. Women were eligible if they resided in the study county, received cash assistance in February of 1997, were single and a U.S. citizen between the ages of 18 and 54, and claimed a racial identity of White or African American (there were too few other minority residents of this county to conduct reliable analyses). A proportional stratified sampling scheme was used from an ordered list of eligible single mothers. To derive a representative sample of the metropolitan area and the study population, cases were proportionately selected by zip code, race (African American or non-Hispanic White), and age. The response rate was 86.2% at Wave 1, 92% at Wave 2, 91% at Wave 3, 90% at Wave 4, and 91% at Wave 5. About half the respondents were African American, 26.6% were aged 35 years or older, 36.0% had three or more children, and 29.5% had not completed high school.

Data from the WES reflect the policy environment in Michigan and in many other states. For example, women in Michigan who work part-time at jobs in which they received the federal minimum wage were at the median for monthly net income for similarly employed residents in 12 states that contain a large portion of the nation’s population and about half of the 1998 Temporary Assistance for Needy Families (TANF) caseload (Acs & Loprest, 2001a). Furthermore, the proportion of the sample that is employed and the proportion that has left welfare are very similar to the corresponding proportions in results of a Manpower Demonstration Research Corporation report on Cleveland (Brock et al., 2002). These findings also parallel results from a Wisconsin study by Maria Cancian, Haveman, Meyer, and Wolfe (2002), as well as those from a Washington, DC, study by Gregory Acs and Pamela Loprest (2001b). Although the current study only uses data from Michigan, the policy and economic conditions faced by welfare recipients in Michigan are broadly representative of those faced by the majority of the U.S. TANF caseload.

Measures

There is no established set of measures to gauge material hardship (Ouellette et al., 2004). Building on the findings of Mayer and Jencks (1989), this article examines changes in four forms of material well-being: food insufficiency status, utility disconnection, unmet medical needs, and housing problems. The food insufficiency measure has been validated by Basiotis (1992) and Cristofar and Basiotis (1992). The other three measures were used by Mayer and Jencks (1989) and have been adopted by a wide variety of surveys, such as the Adult Well-being Topic Module of the Survey of Income and Program Participation (SIPP) and the Fragile Families and Child Well-being Study. Each of these four measures is coded as a dichotomous variable.

Food insufficiency

In each of the five waves, food insufficiency is measured using the single item, “Which of the following statements best describes the food eaten in your household in the last 12 months: enough to eat, sometimes not enough to eat, or often not enough to eat?” Those answering sometimes or often not enough to eat are coded as food insufficient.

Utilities disconnection

This hardship is measured in all five waves. Respondents were asked if their gas or electricity had been turned off because they could not afford to pay the bill since the prior interview (within the last 12 months for Wave 1).

Unmet medical need

This hardship is measured at three waves. Respondents were asked if there was a time that they needed to see a doctor or dentist but could not afford to go since the prior interview.

Housing problems

The presence of this hardship was assessed at three waves. Respondents were asked if they had experienced eight housing upkeep problems in the 12 months prior to the interview—leaky roof or ceiling, plumbing problems, rodents or insects, broken windows, broken heating system, electrical problems, broken stove or refrigerator, inadequate garbage pickup. In keeping with Mayer and Jencks (1989), those who reported three or more difficulties are coded as having housing problems.

As a result of the lack of nationally representative, longitudinal data on material hardship, little is known about the dynamics of material hardship. The explanatory variables to be examined here include controls for household composition, household income, and measures of physical and mental health. Table 1 contains descriptive statistics for the full sample and shows how characteristics for those experiencing each of the hardship types vary relative to the full sample of women on welfare.¹¹

Table 1.

Descriptive Statistics by Material Hardship Type.

	Sample		Food Insufficiency		Utilities Disconnection		Unmet Medical Needs		Housing Problems
	M	SD	M	SD	M	SD	M	SD	M	SD
Household composition
Couple formed	0.130	0.337	0.153	0.360	0.164	0.371	0.084	0.278	0.089	0.284
Couple breaks up	0.091	0.287	0.109	0.312	0.125	0.332	0.090	0.286	0.079	0.270
Total number of adults in household	1.588	0.721	1.545	0.721	1.553	0.688	1.665	0.706	1.596	0.676
Total number of children in household	2.252	1.340	2.402	1.572	2.405	1.565	1.927	1.327	2.505	1.546
Health
Fair/poor health	0.306	0.461	0.449	0.498***	0.370	0.484***	0.443	0.497***	0.426	0.495***
Depression	0.205	0.404	0.338	0.473***	0.312	0.464***	0.255	0.436***	0.289	0.454***
Alcohol dependence	0.022	0.146	0.043	0.205***	0.045	0.208***	0.033	0.179*	0.041	0.198***
Drug dependence	0.029	0.166	0.069	0.253***	0.064	0.246***	0.053	0.225***	0.065	0.246***
Post-traumatic stress disorder	0.150	0.357	0.258	0.438***	0.261	0.440***	0.180	0.385**	0.242	0.429***
Domestic violence	0.137	0.344	0.225	0.418***	0.226	0.419***	0.140	0.347	0.194	0.396***
Gross annual household income	13,484	11678	13,694	221***	12,796	705	14,450	507	12,694	.502**
Observed entries into hardship			141		131		57		36
Observed exits from hardship			92		34		34		21
Total observations with hardship			586		248		369		298

Note: Significance level indicates difference from full sample for those experiencing each hardship.

Significance at .10. **Significance at .05. ***Significance at .01.

In terms of household composition changes, we treat marriages and cohabitating unions jointly.¹² We control for entrances into and the dissolution of unions with separate variables, however. Additionally, we also control for the total number of children and the total number of adults in the households. Our income variable is generated from a measure of gross annual household income. For more information on this measure and how it correlates with monthly measures of income in this sample, see Danziger, Heflin, Corcoran, Oltmans, and Wang (2002).

Physical health is assessed using a single-item measure of self-rated health, which has been shown to be a reliable and valid predictor of mortality and morbidity when controlling for other health status indicators (Idler & Benyamini, 1997; Idler & Kasl, 1995; Shadbolt, 1997). We transform the standard survey 5-item question into a dichotomy indicating if women report being in “fair” or “poor” health (vs. “good,” “very good,” or “excellent”). Domestic violence is indicated by the Conflict Tactic Score, a widely used measure of family violence (Straus & Gelles, 1986). The item indicates recent (within the past 12 months) severe physical abuse. This subscale indicated whether the respondent has been hit with a fist or object, beaten, choked, threatened with a weapon, or forced into sexual activity against her will.¹³

We examine the impact of major depression, post-traumatic stress disorder, alcohol dependence, and drug dependence. All four mental health measures are assessed using the 12-month screening version of the World Health Organization’s (WHO) Composite International Diagnostic Interview (CIDI) at each of the five interviews (Kessler, Andrews, Morczek, Ustun, & Wittchen, 1998; WHO, 1990). The CIDI is a structured interview schedule designed to be used by trained interviewers who are nonclinicians to assess the prevalence of specific psychiatric disorders (Robins, Wing, Wittchen, & Helzer, 1988). WHO field trials and other methodological studies (Wittchen, 1994) have documented acceptable test–retest reliability and clinical validity of the CIDI diagnoses.

Empirical Model

Prior research on material hardship has suffered from two main problems. First, most of the research on material hardship is cross-sectional in nature. Models focus on investigating differences in characteristics in states of material hardship (absence or presence of material hardship, which is measured by a dummy variable that is 0 or 1) using limited dependent variables models, most frequently logistic regression. These models have generated a body of research that indicates the association between material hardship and other individual and household characteristics. The cross-sectional nature of the data is a significant limitation, however, because it both prevents the temporal ordering of the material hardship and the time varying characteristics, such as health problems. Findings from cross-section analysis allow the possibility that both the hardship conditions and the time-varying characteristics are being influenced by factor(s) external to the model, which may include personal characteristics, economic conditions, or social factors.

A more limited number of studies have explored material hardship using time series models (Gundersen & Gruber, 2001; Heflin et al., 2007; Kalil et al., 2002; Reichman et al., 2005). These studies most often use fixed-effect analysis to control for time invariant unmeasured heterogeneity and thus suffer from another limitation. Models that include fixed effects treat entries into hardship and exits from hardship as equivalent, essentially producing the average estimate between the covariates and both transitions in hardship. Conceptually, it might appear odd to argue that the social processes behind exits and entries into material hardship are uniform. Indeed, this would be unthinkable for other common statuses of interest to social stratification researchers, such as unemployment, poverty, or female headship, all of which induce changes to the life course, generally making transitions out of the status more difficult. Yet, in essence, users of fixed-effect models rely on the equivalence of a change from 0 to 1 and from 1 to 0 in order to bring power to their models. This is statistically powerful information, but just as other assumptions sometimes made in statistics, this assumption can be false and should be avoided if possible.

To move the literature forward in this area, we use longitudinal data in order to temporally order our covariates and our outcome variables as well as control for time invariant unobserved heterogeneity. However, unlike prior research, we separately examine transitions into hardship and exits from hardship. Given our small sample sizes, however, we estimate fixed-effect linear probability models (LPMs)¹⁴ instead of logistic regression models. Fixed-effects models assume that there is a fixed, individual component of behavior, social psychology, or utility that affects the outcome being studied, here material hardship. The logic of the model works well for continuous outcomes, where the high, low, or medium level of the outcome is attributed in part to this fixed effect. However, the estimation of a limited dependent variable with logit (or probit) in the context of a small sample size becomes problematic when modeled with fixed effects. Conceptually, one is arguing that there is an underlying propensity for an event to occur (e.g., to experience material hardship). This implies that in the absence of variation in the discrete outcome, which is not difficult with only three or five measures being taken, the fixed effect is infinite and the sample greatly reduced. Infinite fixed effects are not reasonable in theory or life—few people are that dysfunctional—but are unavoidable with probit or logit. Our solution is to use LPM. These models take the form:

Prob (Y_{i t} = 1) = γ_{0} + γ_{1} {H S D}_{i t} + γ_{2} {H L T H}_{i t} + γ_{3} X_{i t} + γ_{4} {I N C}_{i t} + η_{i}

where Y_it is the transition in material hardship status of individual i at time t. We begin by estimating models of the probability of entering material hardship at t for each of the four hardship types for all those were not observed to experience material hardship at t − 1.¹⁵ Then, conditional on reporting hardship at t − 1, we estimate the probability of exiting hardship at time t. HSD_i is a vector of variables capturing household composition, such as the number of children in the household, the number of other adults in the household, the formation of a marriage or cohabitating union, or the dissolution of the same. HLTH_i is a vector of variables capturing physical and mental health. We include a measure of fair or poor health, major depression, post-traumatic stress disorder, drug dependence, and alcohol dependence, as well as domestic violence. X_i is a vector of demographic variables, such as age and race. INC_i is a measure of total gross annual household income. Finally, η is the individual fixed effect that is used to control time invariant unobserved heterogeneity.¹⁶

In additional to fixed-effect LPMs, we also present results for between-effect LPMs. Whereas the fixed-effect models reveal the within-person effects of covariates on material hardship, the between-effect models estimate the effect of covariates across individuals at each time period. In cases where the individual fixed effect dominates the model leaving all the individual covariates insignificant, the between-effects models can give some indication about the individual fixed effect. The benefit of this approach is that the between-effect models can help identify factors that are correlated with the individual fixed effect. The cost of this approach is that unmeasured but fixed factors, such as risk taking, future planning, or community effects, can cause both the fixed effect in material hardship and the fixed explanatory variable. This problem is invariably present in cross-sectional analyses.

Results

In this section, we separately model entry into and exit from four types of material hardship—food insufficiency, utilities disconnection, unmet medical need, and housing problems. For each outcome, cases contribute to the estimation only if they are at risk for experiencing the change in status. Specifically, only cases that are not observed in material hardship at t − 1 contribute to the estimation of models of entry into hardship at time t. Similarly, only cases that are observed to be in material hardship at t − 1 contribute to the estimation of models of exit from material hardship at time t. We begin with fixed-effect models, which estimate within-person effects of each covariate on hardship, and then discuss between-effect estimates, which estimate across person within time period estimates of each covariate on hardship.

Models of Entry Into Material Hardship

Entry into food insufficiency appears to be largely driven by factors that are external to the model. Results presented in Table 2 show that none of the covariates examined are associated with entry into material hardship, suggesting that the individual fixed effect is the most powerful predictor. Individuals who experience food insufficiency are different from those who are not. This difference, termed the individual fixed effect, is more important than over-time differences in health, household composition, and income in terms of predicting entry into food insufficiency. However, between-effects results from Table 3 indicate that women who enter food insufficiency have more children and fewer adults in the household. Additionally, women are more likely to be in fair or poor health themselves and suffer from post-traumatic stress disorder and domestic violence. Finally, women who enter food insufficiency are likely to have lower gross annual household incomes than those who remain food sufficient. The between effects refer to average income over time, or permanent income as it is called in economics, rather than changes in income, or transitory income as it is called in economics. Taken together, the fixed-effect and between-effect results indicate that persistently low income appears to have a stronger association than annual changes in income with food insufficiency.

Table 2.

Fixed-Effect Models of Entry Into Material Hardship.

	Food Insufficiency		Utilities Disconnection		Unmet Medical Needs		Housing Problems
	M	SD	M	SD	M	SD	M	SD
Household composition
Couple breaks up	−0.018	0.029	−0.021	0.023	−0.020	0.070	0.064	0.051
Total number of adults in household	0.003	0.015	−0.014	0.013	−0.026	0.037	−0.014	0.029
Total number of children in household	0.009	0.012	0.015	0.009	−0.007	0.031	0.011	0.021
Health
Fair/poor health	0.014	0.024	0.023	0.018	0.042	0.056	0.063	0.040
Depression	−0.023	0.029	−0.001	0.019	−0.026	0.057	0.060	0.044
Alcohol dependence	−0.139	0.086	0.010	0.052	0.337**	0.157	−0.479***	0.136
Drug dependence	0.090	0.077	0.090**	0.045	−0.092	0.155	−0.039	0.128
Post-traumatic stress disorder	0.015	0.029	−0.024	0.022	−0.038	0.062	0.143***	0.047
Domestic violence	−0.023	0.029	−0.038*	0.022	0.141*	0.072	0.088	0.054
Gross annual household income	0.000	0.008	0.010	0.007	0.058***	0.019	0.039***	0.012
No health insurance					0.083	0.082
Number of observations	1,820		2,127		751		799

Note: The population at risk for moving into hardship at time t is the population not in hardship at t − 1. Each household can contribute multiple observations.

Significance at .10. **Significance at .05. ***Significance at .01.

Table 3.

Between-Effect Models of Entry Into Material Hardship.

	Food Insufficiency		Utilities Disconnection		Unmet Medical Needs		Housing Problems
	M	SD	M	SD	M	SD	M	SD
Household composition
Couple breaks up	−0.021	0.057	0.062	0.040	−0.086	0.102	−0.039	0.087
Total number of adults in household	−0.041*	0.023	−0.013	0.016	0.020	0.032	0.000	0.028
Total number of children in household	0.0214**	0.010	0.015**	0.007	−0.025*	0.014	.025*	0.013
Health
Fair/poor health	0.072**	0.036	0.028	0.026	0.054	0.050	0.052	0.044
Depression	−0.001	0.047	.091***	0.033	−0.007	0.064	0.045	0.057
Alcohol dependence	0.097	0.107	−0.015	0.083	−0.165	0.176	−0.069	0.163
Drug dependence	0.008	0.106	−0.013	0.083	0.022	0.166	−0.143	0.174
Post-traumatic stress disorder	0.129**	0.053	−0.008	0.004	0.050	0.068	0.116*	0.064
Domestic violence	0.102**	0.051	0.0746**	0.036	0.086	0.070	0.095	0.069
Gross annual household income	−0.071***	0.018	−0.042***	0.013	−0.020	0.018	−0.033**	0.017
No health insurance					0.1988***
Number of observations	1,820		2,127		751		799

Note: The population at risk for moving into hardship at time t is the population not in hardship at t − 1. Each household can contribute multiple observations.

Significance at .10. **Significance at .05. ***Significance at .01.

Moving on to models of utility disconnection, fixed-effect results (Table 2) suggest that women are most likely to have utility disconnections after periods of heavy drug use. Drug dependence, however, is quite rare with only about 3% of the sample meeting the criteria at any observation, suggesting that finding is more about the severe consequences of heavy drug use and less about the high prevalence among the welfare sample. Periods of domestic violence, however, are associated with periods of lower odds of utility disconnection. One possible explanation for this finding is that assistance for domestic violence (such as moving to a residential treatment facility) may also provide a buffer against other forms of material hardship, such as utility disconnections. Between-effects models (Table 3), which once again estimate differences in the population who experience utility disconnections and those who do not, indicate households who enter into utility disconnections are likely to have more children present than those who do not experience utilities disconnections. Additionally, both maternal depression and domestic violence distinguish households who have their utilities disconnected from those who do not. The inconsistency of the findings between the between-effect and fixed-effects models suggests that unmeasured characteristics associated with maternal depression and domestic violence are positively associated with utility disconnections. Finally, similar to food insufficiency, households who enter into a period of utility disconnection have lower average incomes than those who do not. Once again, this suggests that unmeasured characteristics associated with long-term income may be more strongly related to the risk of entering into a utility disconnection than a short-term change in income.

Fixed-effect models of entry into having an unmet medical need indicate that for some women periods of alcohol dependence may precede periods of medical need. Similar to drug dependence, only about 3% of the sample meets the criteria for alcohol dependence at any point in time. Once again, domestic violence appears to be a serious risk factor. Specifically, the onset of domestic violence is associated with an increased risk of having an unmet medical need. Interestingly, women were more likely to enter into medical hardships after periods of higher income. One possible interpretation for this finding is that households may lose public health insurance coverage as their earnings and income levels rise. Between-effect models indicate that differences across women associated with the beginning of a period of having an unmet medical need were found in terms of household composition. Women with more children were less likely to enter into medical hardships, perhaps an indication that Medicaid coverage is more likely to be obtained by those with more children. Additionally, while not having health insurance was not a significant predictor in the fixed-effect models, health insurance is a strong predictor of entry into medical hardships in between-effects models. This suggests that women who have health insurance and those who do not differ in terms of unmeasured characteristics that are associated with entry into medical hardships. Furthermore, unmeasured characteristics associated with not having health insurance are stronger predictors of having an unmet medical need than the loss of health insurance coverage.

Finally, we consider entry into housing hardships. From the fixed-effect models, we observe that women are less likely to enter into housing hardships after periods of alcohol dependency. Post-traumatic stress disorder is also associated with a higher probability of entering housing hardships in the next period. Women are also more likely to report entering housing hardships after periods of higher household income, which could result from the loss of subsidized housing, similar to the argument above that unstable higher income is associated with unmet medical needs. Note that the positive sign could also be interpreted in the opposite direction, such that a fall in income reduces entrances into housing hardship. In either case, there is evidence that high implicit welfare tax rates are associated with rising incomes, such that aggregate household well-being may suffer when income increases are offset by reductions or losses of other social benefits (Edin & Lein, 1997; Morgan, Acker, & Weigt, 2010). The between-effect models (Table 3), which indicate differences between households that experience housing hardship and those that do not, suggest that women who enter into housing hardships tend to have lower permanent household incomes than women who do not. This again suggests that women with low incomes are very sensitive to the loss or reduction of benefits associated with rising incomes. As with food and utility hardships, households with more children are more likely to enter housing hardships than households with fewer children. This presents an important screening device and also implies that when these hardships are experienced, many children are being affected. Finally, households who enter into housing hardships are more likely to have a mother who meets the criteria for post-traumatic stress disorder than those who do not, suggesting that unobserved characteristics associated with maternal mental health are related to the risk of housing hardship.

Estimating Exits From Material Hardship

Models of exits from each of the four hardship conditions tend to be weaker than those presented above for entry into hardship. Partially, this is due to the smaller sample size for each of the models, since only those who experience the hardship are included as being at risk of exiting the hardship. Partially, however, this is also due to the difficulty in modeling the social processes that contribute to a household who is in disequilibrium being able to right itself and meet basic necessities again.

As with entries into hardship, we begin by examining exits from food hardships. Fixed-effects estimates are presented in Table 4. As with entries into food hardship, exits from food hardship are largely driven by individual differences across women and not by over-time differences within households. The one exception to this statement is that as the number of children in the household increases, perhaps because of a birth or the addition of a subfamily into the household, the probability of exiting food hardship increases. Interestingly, this same effect is found in the between-effects estimates (Table 5), suggesting that households with more children are more likely to exit from food hardships than households with fewer children. Adult have the opposite effect. Households with more adults have lower odds of exiting from food hardships than households with fewer adults. This suggests that among food-insufficient households, an additional adult is more likely to represent an extra mouth to feed instead of bringing additional resources into the household. Poor maternal health, post-traumatic stress disorder, and domestic violence were all associated with higher post-traumatic of exiting food hardships. Perhaps, among those experiencing food hardships, women with poor physical health, mental health conditions, and domestic violence are more likely to be connected to food referral networks than women without similar conditions but who are food insufficient. Similarly, higher household income is associated with a lower probability of exiting food hardships, suggesting that those with higher incomes may find it harder to find and qualify for emergency food assistance. All these effects show bad circumstances associated with recovery from food hardships. Reversing the explanation, households that have food insufficiency without any of these problems are unlikely to recover, perhaps because they are not functioning very well. When other problems bring on the food insufficiency, addressing that problem becomes primary.

Table 4.

Fixed-Effect Models of Exit From Material Hardship.

	Food Insufficiency		Utilities Disconnection		Unmet Medical Needs		Housing Problems
	M	SD	M	SD	M	SD	M	SD
Household composition
Couple breaks up	−0.082	0.097	−0.251	0.261	0.133	0.146	−0.344	0.240
Total number of adults in household	0.001	0.068	−0.027	0.159	0.157	0.118	0.116	0.120
Total number of children in household	0.086	0.039	−0.009	0.110	0.076	0.075	0.099	0.151
Health
Fair/poor health	−0.121	0.092	0.462	0.278	−0.118	0.136	−0.236	0.172
Depression	0.054	0.083	0.140	0.266	−0.013	0.142	−0.059	0.191
Alcohol dependence	−0.070	0.164	0.003	0.455	0.233	0.346	−0.041	0.360
Drug dependence	−0.135	0.148	0.412	0.392	0.462	0.296	−0.186	0.392
Post-traumatic stress disorder	0.075	0.086	0.117	0.222	−0.001	0.131	−0.009	0.163
Domestic violence	0.030	0.090	0.081	0.253	−0.288	0.183	0.021	0.212
Gross annual household income	−0.051	0.048	−0.052	0.175	−0.148	0.055	−0.204	0.089
No health insurance					0.199	0.116
Number of observations	528		219		328		268

Note: The population at risk for exiting hardship at time t is the population in hardship at t − 1. Each household can contribute multiple observations.

Significance at .10. **Significance at .05. ***Significance at .01.

Table 5.

Between-Effect Models of Exit From Material Hardship.

	Food Insufficiency		Utilities Disconnection		Unmet Medical Needs		Housing Problems
	M	SD	M	SD	M	SD	M	SD
Household composition
Couple breaks up	−0.021	0.057	0.007	0.068	0.050	0.115	−0.036	0.105
Total number of adults in household	−0.041*	0.023	−0.015	0.042	0.000	0.046	−0.011	0.050
Total number of children in household	0.021**	0.010	0.029	0.018	−0.031	0.024	0.028	0.021
Health
Fair/poor health	0.072**	0.036	0.072	0.064	0.129*	0.067	−0.031	0.068
Depression	−0.001	0.047	−0.004	0.070	0.125	0.082	−0.062	0.076
Alcohol dependence	0.097	0.108	0.339	0.208	−0.264	0.227	0.279	0.187
Drug dependence	0.008	0.106	0.000	0.128	0.056	0.182	0.148	0.133
Post-traumatic stress disorder	0.129**	0.053	0.222*	0.081	−0.019	0.104	0.131	0.085
Domestic violence	0.102**	0.051	0.002	0.073	0.067	0.102	0.164*	0.087
Gross annual household income	−0.071***	0.018	−0.038	0.038	0.057**	0.026	0.055	0.036
No health insurance					0.065	0.066
Number of observations	528		219		328		268

Note: The population at risk for exiting hardship at time t is the population in hardship at t − 1. Each household can contribute multiple observations.

Significance at .10. **Significance at .05. ***Significance at .01.

Exits from utility hardship are rare events and very difficult to model. Fixed-effect models are unable to identify any within-household characteristics that are associated with a change in the probability of exiting from a utility hardship. Between-effect models do little better with the only significant coefficient being post-traumatic stress disorder: Among women who have their utilities disconnected, those with post-traumatic stress disorder are more likely to have their services restored and exit from utility hardship. Once again, the likely explanation is that women with post-traumatic stress disorder are more likely to be referred to or qualify for services that help them reestablish utility service.

Among households who have experienced an unmet medical need, within-household increases in income are associated with lower probabilities of exiting from medical hardships (Table 4). However, results from the between-effect models indicate that households with higher household incomes are more likely to exit from medical hardships than those with lower incomes. This suggests that the individual fixed effect relating to exiting from medical hardships is correlated with household income. Between-effect models also suggest that among households with unmet medical needs, poor maternal health is associated with an increased probability of exit from medical hardships. One interpretation of this finding is that women with chronic health conditions may be more likely to have their medical needs met than those in better maternal health.

Finally, among those with housing problems, women with higher incomes are less likely to exit from housing problems than they are when they have lower incomes (Table 4). This finding is consistent with the income targeting of housing assistance programs. Between-effect models reveal that women who report domestic violence are more likely to exit from housing hardship than are women who remain in poor housing. One explanation for this finding is that in some localities, women in domestic violence shelters are given priority in public housing programs. Alternatively, domestic violence may lead to residential change, perhaps unintentionally solving the housing quality problem.

Summary

Using data from the WES and methods that address unobserved heterogeneity, we examine determinants of change in material hardship status for four different measures of well-being—food insufficiency, utilities disconnection, unmet medical need, and housing problems. Specifically, we analyze the role of maternal health, household composition, and income on entrances into and exits from material hardship. Our results show that there is a great deal of heterogeneity across forms of material hardship but that in general, the predictions of our conceptual model are confirmed. Determinants of entrances into material hardship differ from those that predict exits, suggesting that interventions to help families exit from material hardship may need to address different issues than those that triggered the entrance into material hardship in the first place. We explore patterns among determinants by hardship below.

Consistent with our conceptual model, we find evidence that maternal health is associated with entry into all four forms of material hardship. The nature of the relationship between maternal health and hardship varies, however. In some cases, the onset (or recurrence) of domestic violence or mental health problems is associated with entry into hardship. In other cases, however, it is not the onset of the condition but the existence of the problem itself, and all the unobserved differences between those with and without the condition, that is associated with a higher risk for material hardship. Our finding that maternal physical and mental health conditions are associated with leaving material hardships is inconsistent with a theory that these problems prolong material hardship. We interpret this finding as an indication that women with more severe difficulties may be more likely to receive services that enable them to return to equilibrium, relative to other women without health conditions who report material hardship. It is important to note that households already in a state of material hardship are motivated to solve the problem. Perhaps something similar to domestic violence or post-traumatic stress disorder is tolerated until it results in an inability to meet basic needs. Finally, bad management, or material hardship with no discernible cause, might be difficult to address. Social service agencies finding women in material hardship for no clear reason might be less able to find effective ways to solve the household’s underlying problems.

In terms of household composition, households with more children have a higher probability of reporting food insufficiency, utility disconnections, and housing problems and a lower probability of reporting having an unmet medical need. More children in the household in several cases encourage recovery. This is surely in part a sense that when children have begun to suffer material hardships something must be done. It is important to be clear that it is not the relative increase but the absolute number of children that matters and that the risk may not be attached to the children themselves but unmeasured differences between households of different sizes. Regardless of the source of the variation, the number of children in the household is an easily identifiable risk factor that can be used for screening purposes. If household composition is used as a screening device, however, it should be noted that among food-insufficient households, those with more children (and fewer adults) are more likely to exit from food insufficiency.

Our results for income are complex. Short-term increases in income are noted to increase the probability of entering into housing or medical hardship, perhaps as people become ineligible for income targeted programs because of the high marginal tax rate of social assistance programs. However, those with low permanent incomes face higher risk of food insufficiency, utilities disconnections, and housing problems. One explanation is that low-income households may exhaust their savings and social support networks over time, making it difficult to cover their basic needs. We find little evidence that short-term increases in income help pull households out of hardship except in the case of having an unmet medical need. In fact, in the case of food hardships, household income is negatively related to the probability of exiting. Thus, income-based approaches to address material hardship, at least within the current structure of social programs, are not likely to be effective. This quantitative finding is consistent with qualitative reports that women lose access to social programs when earnings increase and suggests that household well-being needs to be carefully monitored as women transition from welfare to work (Collins & Mayer, 2010; Edin & Lein, 1997; Hays, 2003; Morgan et al., 2010).

There are limitations of this work that should be noted. These results are not generalizable outside of the welfare population. Furthermore, since they are from a single county in Michigan, they may not be considered to be representative of the national welfare sample.¹⁷ Additionally, two of the four measures are only measured at three instead of five time points. This limits the statistical power of the model. Finally, this work considers ways in which risk factors may affect material hardship. However, it is possible that causation goes the other way—that material hardship may affect the risk factors. Note, however, that all changes in material hardship are timed after the risk factors, never contemporaneous in the same interview. Causation could still be reversed, if, for example, impending problems lead to stress-related changes in health, but this research avoids the problems caused by measuring medical need and change in health status at the same time.

Discussion

In general, our models do a better job predicting entrances into material hardship than exits. There is very little research using either survey or ethnographic data on how women who are experiencing material hardship are able to cobble together the resources to exit hardship and return to a state of well-being. Two ethnographic studies that examined survival strategies (Edin & Lein, 1997; Hill & Kauff, 2001) identify a number of work-related strategies, network-based strategies, and agency-based strategies that women pursue to ensure that their basic needs are met. However, our results indicated that different hardships have different triggers and that the social processes in play for different forms of material hardship may vary. This conclusion is echoed by more recent studies (Heflin et al., 2009; Heflin et al., 2011). It is clear that additional research is needed to explore how families make trade offs between different necessities and to further our understanding of the different processes that families employ to return to equilibrium.

Understanding the triggers associated with entrance and exit from material hardship is an issue of great social significance, given the growing evidence regarding the consequences of experiencing material hardship for both adult and child health outcomes. For children, research indicates that material hardship has a significant negative impact on child development (Gershoff et al., 2007). For adults, there is a body of research indicating that mental health is negatively affected by periods of experiencing material hardship (Casey et al., 2004; Heflin, Siefert, Corcoran, & Williams, 2005; Laraia, Siega-Riz, Gundersen, & Dole, 2006). Policy simulations by Heflin and Iceland (2009) suggest that larger improvements in health could be achieved by meeting basic needs such as food, adequate housing, and utilities than by increasing income alone.

Results from this research, particularly the nuanced relationship between income and household well-being, provides additional support for the well-established critique of the current income poverty measure and the value of examining alternative measures of well-being, such as the direct measures of the ability to provide necessities examined here. It is well established in the empirical literature at this point that the association between income poverty and various hardship measures indicates that they are only moderately correlated with one another in the United States (Beverly, 2000; Boushey et al., 2001; Bradshaw & Finch, 2003; Mayer, 1997; Mayer & Jencks, 1989, 1993; Perry, 2002; Rector et al., 1999). Yet participation in federal social welfare programs continues to be tied to a multiplier of the federal poverty threshold. Evidence suggests that programs need to be available to those with incomes up to at least 200% of the poverty line as material hardship is not limited to those living in poverty. Raising the income eligibility cutoff would also help address the increased short-term risk of material hardship reported as incomes rise as a result of the high marginal tax rate of social welfare program.

Finally, there are several risk factors that could be used in policy practice in screening for material hardship. Many social welfare programs require information on the number of children that could be used to identify those at higher risk of material hardship. Additionally, the importance of maternal health suggests that when routine screenings indicate problems with maternal physical or mental health, as well as domestic violence, crisis interventions to support family well-being may be needed. An important, although well known, policy implication is that in most cases, preventing material hardship is more effective than reacting to the problem later.

Footnotes

Authors’ Note

Errors remain the responsibility of the authors alone.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article:

Funding for this research was provided through the small grant program at the Institute for Research on Poverty at the University of Wisconsin and Economic Research Service, U.S. Department of Agriculture. Funding for the Women’s Employment Survey was provided by grants from the Charles Stewart Mott Foundation, the Joyce Foundation, the John D. and Catherine T. MacArthur Foundation, the Substance Abuse Policy Research Program of the Robert Wood Johnson Foundation, the National Institute of Child Health and Human Development, and the National Institute of Mental Health (R24-MH51363).

Notes

References

Acs

Loprest

(2001a). Final synthesis report of findings from ASPE’s leavers grants. Report to the U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation (Contract No. HHS-100-99-0003). Washington, DC: Urban Institute.

Acs

Loprest

(2001b). The status of TANF leavers in the District of Columbia: Final report. Washington, DC: Urban Institute.

Alaimo

Briefel

R. R.

Frongillo

E. A.

Olson

C. M.

(1998). Food insufficiency in the United States: Results from the Third National Health and Nutrition Examination Survey (NHANES III). American Journal of Public Health, 88, 419-426.

Basiotis

P. P.

(1992). Validity of the self-reported food sufficiency status item in the US Department of Agriculture Food Consumption Surveys. Presented at the proceedings of the annual meeting of the American Council in the Consumer Interest, Toronto, Ontario, Canada.

Beverly

S. G.

(2001). Measures of material hardship: Rationale and recommendation. Journal of Poverty, 5, 23-41.

Beverly

S. G.

(2000). Using measures of material hardship to assess well-being. Focus, 21(2), 65-69.

Bhargava

Jolliffe

Howard

(2008). Socioeconomic, behavioral and environmental factors predicted body weights and household food insecurity scores in the Early Childhood Longitudinal Study–Kindergarten. British Journal of Nutrition, 100, 438-444.

Bitler

Gundersen

Marquis

(2005). Are WIC non-recipients at less nutritional risk than recipients? An application of the food security measure. Review of Agricultural Economics, 27, 433-438.

Blank

(2007). Presidential address: How to improve poverty measurement in the United States. Journal of Policy Analysis and Management, 27, 233-254.

10.

Boushey

Gundersen

(2001). When work just isn’t enough: Measuring hardship faced by families after moving from welfare to work (Briefing Paper). Washington, DC: Economic Policy Institute.

11.

Bradshaw

Finch

(2003). Overlaps in dimensions of poverty. Journal of Social Policy, 32, 513-525.

12.

Brock

Coulton

London

Polit

Richburg-Hayes

Scott

. . . Seith

(2002). Welfare reform in Cleveland: Implementation, effects, and experiences of poor families and neighborhoods (Report, Project on Devolution and Urban Change). New York, NY: Manpower Demonstration Research Corporation.

13.

Bronte-Tinkew

Zaslow

Capps

Horowitz

McNamara

(2007). Food insecurity works through depression, parenting, and infant feeding to influence overweight and health in toddlers. Journal of Nutrition, 137, 2160-2165.

14.

Cancian

Haveman

Meyer

D. R.

Wolfe

(2002). Before and after TANF: The economic well-being of women leaving welfare (Special Rep. No. 77). Madison: University of Wisconsin–Madison.

15.

Casey

Goolsby

Berkowitz

Frank

Cook

Cutts

. . . Children’s Sentinel Nutritional Assessment Program Study Group. (2004). Maternal depression, changing public assistance, food security, and child health status. Pediatrics, 113, 298-304.

16.

Citro

C. F.

Michael

R. T.

(1995). Measuring poverty: A new approach. Washington, DC: National Academies Press.

17.

Collins

J. L.

Mayer

(2010). Both hands tied: Welfare reform and the race to the bottom of the low-wage labor-market. Chicago, IL: University of Chicago Press.

18.

Cristofar

S. P.

Basiotis

P. P.

(1992). Dietary intakes and selected characteristics of women ages 19-50 and their children ages 1-5 years by reported perceptions of food insufficiency. Journal of Nutrition Education, 24(2), 53-58.

19.

Danziger

Heflin

C. M.

Corcoran

M. E.

Oltmans

Wang

H-C.

(2002). Does it pay to move from welfare to work? Journal of Policy Analysis and Management, 21, 671-692.

20.

Ecob

Smith

G. D.

(1999). Income and health: What is the nature of the relationship? Social Science & Medicine, 48, 693-705.

21.

Edin

Lein

(1997). Making ends meet: How single mothers survive welfare and low-wage work. New York, NY: Russell Sage.

22.

Gershoff

E. T.

Aber

J. L.

Raver

C. C.

Lennon

M. C.

(2007). Income is not enough: Incorporating material hardship into models of income associations with parenting and child development. Child Development, 78, 70-95.

23.

Gundersen

Gruber

(2001). The dynamic determinants of food insufficiency. In Andrews

Prell

(Eds.), Second food security measurement and research conference (pp. 91-109, Volume II: Papers. Food Assistance and Nutrition Rep. No. 11-2). Washington, DC: U.S. Department of Agriculture, Economic Research Service.

24.

Gundersen

Lohman

Garasky

Stewart

Eisenmann

(2008). Food security, maternal stressors, and overweight among low-income U.S. children: Results from NHANES 1999-2002. Pediatrics, 122, e529-e540.

25.

Hamilton

W. L.

Cook

J. T.

Thompson

W. W.

Buron

L. F.

Frongillo

E. A.

Olson

C. M.

Wehler

C. A.

(1997). Household food security in the United States in 1995 (Technical Report of the Food Security Measurement Project). Washington, DC: Food and Consumer Service, U.S. Department of Agriculture.

26.

Hays

(2003). Flat broke with children: Women in the age of welfare reform. New York, NY: Oxford University Press.

27.

Heflin

(2006). Dynamics of different forms of material hardship in the women’s employment survey. Social Service Review, 80, 377-397.

28.

Heflin

London

Scott

(2011). Mitigating material hardship: The strategies low-income mothers employ to reduce the consequences of poverty. Sociological Inquiry, 81(2), 1-24.

29.

Heflin

C. M.

Corcoran

M. E.

Siefert

(2007). Work trajectories, income changes, and food insufficiency in a Michigan welfare population. Social Service Review, 81, 3-25.

30.

Heflin

C. M.

Iceland

(2009). Poverty, hardship and depression. Social Science Quarterly, 90, 1051-1071.

31.

Heflin

C. M.

Sandberg

Rafail

(2009). The structure of material hardship in U.S. households: An examination of the coherence behind common measures of well-being. Social Problems, 56, 746-764.

32.

Heflin

Siefert

Corcoran

M. E.

Williams

D. R.

(2005). Food insufficiency and the mental health of current and recent welfare recipients: Findings from a longitudinal survey. Social Science & Medicine, 61, 1971-1982.

33.

Hill

Kauff

(2001). Living on little: Case studies of Iowa families with very low income. Washington, DC: Mathematica Policy Research.

34.

Iceland

Bauman

(2004). Income poverty and material hardship: How strong is the association? (National Poverty Center Working Paper Series No. 04-17). Ann Arbor: Gerald R. Ford School of Public Policy, University of Michigan.

35.

Idler

E. L.

Benyamini

(1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior, 38, 21-37.

36.

Idler

E. L.

Kasl

S. V.

(1995). Self-ratings of health: Do they also predict change in functional ability? Journal of Gerontology: Social Sciences, 50B, S344-S353.

37.

Kalil

Seefeldt

K. S.

Wang

H.-C.

(2002). Sanctions and material hardship under TANF. Social Service Review, 76, 642-662.

38.

Kessler

R. C.

Andrews

Morczek

Ustun

Wittchen

H. U.

(1998). The World Health Organization Composite International Diagnostic Interview Short-Form (CIDI-SF). International Journal of Methods in Psychiatric Research, 7, 171-185.

39.

Laraia

B. A.

Siega-Riz

A. M.

Gundersen

Dole

(2006). Psychological factors and socioeconomic indicators are associated with household food insecurity among pregnant women. Journal of Nutrition, 136, 177-182.

40.

Link

Phelan

J. C.

(1995). Social conditions as fundamental causes of disease. Journal of Health and Social Behavior, Extra Issue, 80-94.

41.

Lynch

Smith

G. D.

Harper

Hillemeier

Ross

Kaplan

Wolfson

(2004). Is income inequality a determinant of population health? Part 1. A systematic review. Millbank Quarterly, 82(1), 5-99.

42.

Mayer

(1997). What money can’t buy: Family income and children’s life chances. Cambridge, MA: Harvard University Press.

43.

Mayer

Jencks

(1989). Poverty and the distribution of material hardship. Journal of Human Resources, 24, 88-114.

44.

Mayer

S. E.

Jencks

(1993). Recent trends in economic inequality in the United States: Income versus expenditures versus material well-being. In Wolff

E. N.

(Ed.), Poverty and prosperity in the USA in the late twentieth century (pp. 121-203). New York, NY: St. Martin’s Press.

45.

Morgan

Acker

Weigt

(2010). Stretched thin: Poor families, welfare work, and welfare reform. Ithaca, NY: Cornell University Press.

46.

Nelson

Brown

M. E.

Lurie

(1998). Hunger in an adult population. Journal of the American Medical Association, 279, 1211-1214.

47.

Ouellete

Burstein

Long

Beecroft

(2004). Measuring material hardship (Final report). Washington, DC: Office of the Assistant Secretary for Planning and Evaluation, Office of the Secretary, U.S. Department of Health and Human Services.

48.

Perry

(2002). The mismatch between income measures and direct outcome measures of poverty. Social Policy Journal of New Zealand, 19, 101-127.

49.

Polit

D. F.

London

A. S.

Martinez

J. M.

(2000). Food security and hunger in poor, mother-headed families in four U.S. cities. New York, NY: Manpower Demonstration Research Corporation.

50.

Rector

(1999, September 18). The myth of widespread American poverty. Washington, DC: Heritage Foundation.

51.

Reichman

N. E.

Teitler

J. O.

Curtis . (2005). TANF sanctioning and hardship. Social Service Review, 79, 215-236.

52.

Robins

Wing

Wittchen

Helzer

(1988). The Composite International Diagnostic Interview: An epidemiologic instrument suitable for use in conjunction with different diagnostic systems and in different cultures. Archives of General Psychiatry, 45, 1069-1077.

53.

Sen

(1999). Development as Freedom. New York: Alfred A. Knopf.

54.

Shadbolt

(1997). Some correlates of self-rated health for Australian women. American Journal of Public Health, 87, 951-956.

55.

Short

(2005). Material and financial hardship and income-based poverty measures in the USA. Journal of Social Policy, 34, 21-38.

56.

Siefert

Bowman

P. J.

Heflin

Danziger

Williams

(2000). Social and environmental predictors of maternal depression in current and recent welfare recipients. American Journal of Orthopsychiatry, 70, 510-522.

57.

Straus

M. A.

Gelles

R. J.

(1986). Societal change and change in family violence from 1975 to 1985 as revealed by two national surveys. Journal of Marriage and the Family, 48, 465-479.

58.

Wittchen

H. U.

(1994). Reliability and validity studies of the WHO-Composite International Diagnostic Review (CIDI): A critical review. Journal of Psychiatric Research, 28, 57-84.

59.

World Health Organization. (1990). Composite International Diagnostic Review (CIDI, Version 1.0). Geneva, Switzerland: Author.